More Comprehensive Analysis

Toward filling the gaps in these studies, the author collected data on 181 characters of the dentition (97), cranium (30), and postcranium (54). A detailed description of these characters and a discussion of their distribution is available elsewhere (Silcox, 2001). All the characters that have been used to support Primatomorpha (Beard, 1993a), and Volitantia (Simmons, 1995; Simmons and Geisler, 1998; Simmons and Quinn, 1994), that could be considered for fossils were assessed.

In selecting taxa to study, the author included members of all 11 families of plesiadapiforms known in 2001 (Carpolestidae, Plesiadapidae, Microsyopidae, Paromomyidae, Picromomyidae, Purgatoriidae, Micromomyidae, Toliapinidae, Palaechthonidae, Saxonellidae, and Picrodontidae). Plesiadapiform species to be sampled were selected based on two criteria. First, all the basal-most members of groups for which previous studies are informative were included (i.e., Elphidotarsius and Chronolestes for Carpolestidae—Simpson, 1928, 1935b; Rose, 1975, 1977, Beard and Wang, 1995; Pandemonium and Pronothodectes for Plesiadapidae—Gingerich, 1976; Van Valen, 1994; Navajovius, Niptomomys, and Arctodontomys for Microsyopidae—Gunnell, 1985, 1989; Paromomys for Paromomyidae—Simpson, 1955; Bown and Rose, 1976; and Picromomys for Picromomyidae—Rose and Bown, 1996). For all the other families poor sampling and/or a lack of consensus about the internal relationships of the group mandated the study of all known species (Purgatoriidae, Micromomyidae, Toliapinidae, Palaechthonidae, Saxonellidae, and Picrodontidae). Second, taxa preserving features of the cranium or postcranium were included, even if they are considered to be well nested within their respective families (e.g., Plesiadapis tricuspidens). The end result was a list of 62 species of plesiadapiforms to be studied, representing about half the total number currently recognized. Representatives of Chiroptera, Dermoptera, Scandentia, Mixodectidae, Plagiomenidae, and Euprimates were also chosen for analysis using similar criteria. Primitive eutherians (leptictids and Asioryctes nemegtensis) were employed as outgroups.

Data was initially collected at the species level. Cladistic analyses were run on the three major data partitions (dental, cranial, and postcranial) at the species level using PAUP* 4.0P (Swofford, 2001), and character distributions were studied using MacClade 3 (Maddison and Maddison, 1992). Families whose monophyly (all but Purgatoriidae, Palaechthonidae, and Toliapinidae sensu Hooker et al., 1999) were well supported were then combined. In cases where a family-level grouping could not be used, genera were employed when a genus was supported as monophyletic (i.e., Toliapina). The resulting dataset, using higher taxonomic groupings, was analyzed for each of the three data partitions and in a total evidence analysis (following the reasoning of Kluge, 1989). The discussion here will focus on this total evidence analysis, since this approach allows all for conflicting patterns of character distribution to compete directly. It is worth noting that the conclusions of the partitioned analyses did not always coincide with those arising from the total evidence analysis. Particularly, the postcranial analyses showed support for a Paromomyidae + Volitantia clade, while the cranial analyses indicated (very weak) support for a Microsyopidae + Volitantia grouping. The former conclusion is subject to revision in light of recent discoveries of new plesiadapi-form postcranials (Boyer et al., 2001; Bloch and Boyer 2002, 2003 and 2006; Bloch et al., 2002, 2003), and new descriptions of the extant scandentian Ptilocercus lowii (Sargis, 2002a, b, and 2006), which were not available at the time of data collection. A collaborative project that includes these new data is currently underway (Bloch and Boyer, 2003; Bloch et al., 2002, 2004). The microsyopid/volitantian node is so poorly supported as to be unconvincing in light of the evidence from other parts of the study. The cranial analysis did not uphold Wible and Covert's (1987) claim of basicranial support for a Euprimate-Scandentia clade that excludes plesiadapiforms, or Kay et al.'s (1990, 1992) claim for cranial support of a paromomyid-dermopteran clade that excludes Euprimates (see also Bloch and Silcox, 2001, 2006). In fact, recent discoveries have documented one of the features that Wible and Covert cited as being key to supporting a euprimate-scandentian clade in a plesiadapiform (a bony tube for the internal carotid nerves in Ignacius; Silcox, 2003).

The total evidence analysis is open to criticism in that the results may largely reflect patterns of relationships indicated by the dental data. Many workers seem to consider dental data to be less reliable than other types of information, on the grounds that it supposedly shows more homoplasy than other parts of the skeleton (see discussion in Van Valen, 1994; Silcox, 2001). A quantitative study that actually analyzed the amount of homoplasy in different skeletal systems found no significant differences between dental, cranial, and postcranial regions (Sanchez-Villagra and Williams, 1998). What is more, teeth offer an important advantage over other parts of the skeleton. While for plesiadapiforms in general the only cranial and postcranial remains that are available belong to advanced members of different families, dentitions are known for very primitive as well as very derived forms. This allows one to get closer to the actual branching points, minimizing the confounding effects of interfamilial evolution producing convergences to other derived forms. In other words, study of dental features avoids long branch problems that are likely to be marked in more poorly sampled systems.

A series of heuristic searches totaling 3000 replicates, starting from different random trees and swapping on all starting trees, was performed on the total dataset with monophyletic higher taxa (i.e., families or genera) combined. This dataset included 38 taxa scored for all 181 characters. The search found 20 most parsimonious trees of length = 788 steps, CI(consistency index) = 0.490, RI(retention index) = 0.521, and RC(rescaled consistency index) = 0.255. The strict consensus tree resulting from these 20 trees was largely unresolved as a result of a couple of "wildcard" taxa that occupied very different positions on the various trees (i.e., Mixodectes and Eudaemonema). An effective way of dealing with such wildcard taxa is to calculate an Adams consensus tree. In an Adams consensus tree wildcard taxa appear unresolved at the highest node at which their position can be ascertained, with no loss of resolution "upstream". As such, examining an Adams consensus tree provides a better view of the pattern of relationships suggested by the data, since relationships that are well documented will be retained. The resulting Adams consensus tree is given as Figure 4.

All of the 20 most parsimonious trees include a clade that contains all plesiadapiforms and euprimates, and that excludes Scandentia, Chiroptera, and Dermoptera—this is also reflected in the Adams consensus tree at the node labeled "Primates". Rather than exhibiting a close relationship to Paromomyidae, Dermoptera is part of a monophyletic Volitantia, which may be more closely related to Scandentia than plesiadapiforms and Eupri-mates. Removal of Chiroptera from the analysis (as suggested by molecular results; Miyamoto et al., 2000; Pumo et al., 1998; Waddell et al., 1999; Springer et al., 2006) does not impact the relationships between the remaining taxa (e.g., Primatomorpha and Eudermoptera are not re-formed).

Leptictidae Asioryctes






Mixodectes ■ Eudaemonema cuspidata Purgatorius unio Purgatorius janisae - Berruvius

Microsyopidae ri

Palaechthon alticuspis Torrejonia

Plesiolestes problematicus Palaechthon nacimienti Paromomyidae Premnoides douglassi Palaechthon woodi Anasazia williamsoni Picrodontidae Picromomys petersonorum Pandemonium dis Chronolestes simul Carpolestidae Plesiadapidae Saxonellidae Palenochtha minor Avenius amatorum Toliapina

Palenochtha weissae Altiatlasius koulchii Altanius orlovi Adapidae Omomyidae


Figure 4. Adams consensus tree resulting from 3000 replicates of a heuristic search of the combined dental, cranial, and postcranial datasets (181 total characters) from Silcox, 2001. Twenty most parsimonious trees were found of length =788 steps, CI = 0.490, RI = 0.521, and RC = 0.255. The dotted lines leading to Mixodectes and Eudaemonema indicate the lack of certainty surrounding the relationships of these taxa—they were not included in Primates for this reason. Although Berruvius appears in an unresolved position, this is purely a product of missing data and the available dental evidence indicates a sister group relationship with Microsyopidae.

Plesiadapiformes itself is not a monophyletic taxon to the exclusion of Euprimates. This is a feature of every result (i.e., species-level, family-level, partitioned, and total evidence) found from the various analyses run in this study. The paraphyletic nature of Plesiadapiformes implies a fairly high level of homoplasy in the evolution of the group, as also indicated by the relatively low CI. The structure of this tree implies that some unusual features, such as an I1 with an apical division, evolved more than once. Although most of the clearest evidence of homoplasy rests with the dental data partition, this effect turns out to be a product of effective sample size (Silcox, 2001).

Three families were also not found to be monophyletic-Purgatoriidae, Palaechthonidae, and Toliapinidae sensu Hooker et al., 1999. The former two taxa appear to be generally primitive, paraphyletic clusters. Toliapinidae can be rendered monophyletic if Berruvius is transferred back to the Microsyopidae (where it had generally been considered to reside until Hooker et al., 1999; see Gunnell, 1989; Russell, 1981).

0 0

Post a comment